|
|
|
| Study on Predicting the Surface Settlement for Shield Tunneling Based on DEACO-WNN |
| HAO Ru-jiang1, JI Yan-peng1, NI Zhen-li2 |
1. Shijiazhuang Tiedao University, Shijiazhuang, Hebei 050043, China; 2. The 1st Engineering Co. Ltd of China Railway 22nd Bureau Group, Beijing 100043, China
|
|
|
|
Abstract Abstract：Research purposes：In recent years, the subway career is developing rapidly, a number of new lines were built. Surface settlement caused by shield tunneling brings more trouble for construction safety. A research on reduce security risks during the construction and improve construction quality of surface settlement caused by shield tunneling is significant.
Research conclusions：In this paper, the factors of surface settlement caused by shield tunneling and prediction model were studied, the following conclusions were got:（1）A wavelet neural network (WNN) prediction model was created by soil covering thickness, compression modulus, cohesion, natural density, angle of internal friction, jack thrust, pressure grouting. （2）The surface settlement model was created by the wavelet neural network(WNN),which its original weight, scaling parameters and translation parameters values were obtained by the differential evolution and ant colony optimization algorithm (DEACO) and the goal of optimization was relative entropy. The training speed and prediction accuracy was improved greatly using the DEACO-WNN. （3）The measured data in Beijing 6th subway line were tested in the presented model, and the minimum prediction error was just only 0.5%,the prediction accuracy within the allowed range. （4）The model created in this paper has a good guidance for the security protection area of shield tunneling, it could be helpful to improve the construction technique through effective prediction of surface settlement, and enhancing the safety of construction.
|
|
|
|
|
|
|
|
|
